headers
Amos Shapira | 1 Nov 01:11
Picon

Debian Etch and Xen 3.1 (from Lenny) and XenSource kernels - initial ramdisk problems

Hello,

My hardware is an Intel Xeon 3050 with VT support and 4 Gb RAM.

I previously got Debian Etch and Xen 3.0.3-1 up and running with Windows Server 2003 and CentOS 5 x86_64 DomU's, but in order to support also CentOS 4 and CentOS 3 with 32 bits I got the impression that I need to upgrade to Xen 3.1.

So, trying to follow the process described in http://bderzhavets.blogspot.com/2007/08/xen-3_27.html more or less, I upgraded to Xen 3.1 from Debian Lenny, using kernel and initrd file from XenSource's tar ball (Debian's kernel panics right in the beginning) and I sort of manage to get the CentOS image start to boot but it get stuck on creation of the right device files by the inittial ramdisk.

Here is the current configuration:

kernel = '/boot/vmlinuz-2.6.18-xen'
ramdisk = '/boot/initrd.img-2.6.18-xen'
extra = "ydebug text root=/dev/sda1"
name = "centos50-01"
memory = "512"
disk = [ 'phy:/dev/xen/centos5.0-01-root,ioemu:sda1,w', 'phy:/dev/xen/centos5.0-01-swap,ioemu:sda2,w']
device_model='/usr/lib/xen-3.1-1/bin/qemu-dm'
vif = [ 'type=ioemu', 'bridge=xenbr0', ]
vcpus=1
on_reboot = 'destroy'
on_crash = 'destroy'

And here is the output from "xm create" (it's more verbose because of a "ydebug" kernel command line argument):

Using config file "./centos5.0-01.cfg".
Started domain centos50-01
Bootdata ok (command line is ydebug text root=/dev/sda1)
Linux version 2.6.18-xen (shand <at> endor) (gcc version 3.4.4 20050314 (prerelease) (Debian 3.4.3-13)) #1 SMP Fri May 18 16:01:42 BST 2007
BIOS-provided physical RAM map:
 Xen: 0000000000000000 - 0000000020800000 (usable)
No mptable found.
Built 1 zonelists.  Total pages: 133120
Kernel command line: ydebug text root=/dev/sda1
Initializing CPU#0
PID hash table entries: 4096 (order: 12, 32768 bytes)
Xen reported: 2133.406 MHz processor.
Console: colour dummy device 80x25
Dentry cache hash table entries: 131072 (order: 8, 1048576 bytes)
Inode-cache hash table entries: 65536 (order: 7, 524288 bytes)
Software IO TLB disabled
Memory: 504576k/532480k available (2014k kernel code, 19044k reserved, 874k data, 172k init)
Calibrating delay using timer specific routine.. 4268.62 BogoMIPS (lpj=21343118)Security Framework v1.0.0 initialized
Capability LSM initialized
Mount-cache hash table entries: 256
CPU: L1 I cache: 32K, L1 D cache: 32K
CPU: L2 cache: 2048K
CPU: Physical Processor ID: 0
CPU: Processor Core ID: 0
SMP alternatives: switching to UP code
Freeing SMP alternatives: 24k freed
Brought up 1 CPUs
migration_cost=0
checking if image is initramfs... it is
Freeing initrd memory: 4176k freed
NET: Registered protocol family 16
Brought up 1 CPUs
PCI: setting up Xen PCI frontend stub
ACPI: Interpreter disabled.
Linux Plug and Play Support v0.97 (c) Adam Belay
pnp: PnP ACPI: disabled
xen_mem: Initialising balloon driver.
PCI: System does not support PCI
PCI: System does not support PCI
NET: Registered protocol family 2
IP route cache hash table entries: 32768 (order: 6, 262144 bytes)
TCP established hash table entries: 131072 (order: 9, 2097152 bytes)
TCP bind hash table entries: 65536 (order: 8, 1048576 bytes)
TCP: Hash tables configured (established 131072 bind 65536)
TCP reno registered
IA-32 Microcode Update Driver: v1.14a-xen < tigran <at> veritas.com>
audit: initializing netlink socket (disabled)
audit(1193874811.104:1): initialized
VFS: Disk quotas dquot_6.5.1
Dquot-cache hash table entries: 512 (order 0, 4096 bytes)
Initializing Cryptographic API
io scheduler noop registered
io scheduler anticipatory registered
io scheduler deadline registered
io scheduler cfq registered (default)
rtc: IRQ 8 is not free.
Non-volatile memory driver v1.2
RAMDISK driver initialized: 16 RAM disks of 16384K size 1024 blocksize
loop: loaded (max 8 devices)
Xen virtual console successfully installed as tty1
Event-channel device installed.
netfront: Initialising virtual ethernet driver.
Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
ide: Assuming 50MHz system bus speed for PIO modes; override with idebus=xx
PNP: No PS/2 controller found. Probing ports directly.
i8042.c: No controller found.
mice: PS/2 mouse device common for all mice
md: md driver 0.90.3 MAX_MD_DEVS=256, MD_SB_DISKS=27
md: bitmap version 4.39
NET: Registered protocol family 1
NET: Registered protocol family 17
xen-vbd: registered block device major 8
blkfront: sda1: barriers enabled
blkfront: sda2: barriers enabled
vif vif-0: 2 parsing device/vif/0/mac
netfront: device eth1 has copying receive path.
XENBUS: Timeout connecting to device: device/vif/0 (state 6)
XENBUS: Device with no driver: device/console/0
Freeing unused kernel memory: 172k freed
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/drivers/scsi/scsi_mod.ko
scsi_mod: no version for "struct_module" found: kernel tainted.
SCSI subsystem initialized
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/drivers/scsi/libata.ko
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/drivers/scsi/ata_piix.ko
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/drivers/scsi/sd_mod.ko
register_blkdev: cannot get major 8 for sd
+ mkbdev /dev/sda sda
+ mksymdev /dev/sda /sys/block/sda/dev b
+ devfile=/dev/sda
+ sysfile=/sys/block/sda/dev
+ cb=b
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Waiting 1 seconds for /sys/block/sda/dev to show up
Waiting 1 seconds for /sys/block/sda/dev to show up
+ sleep 1
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Waiting 2 seconds for /sys/block/sda/dev to show up
Waiting 2 seconds for /sys/block/sda/dev to show up
+ sleep 2
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Waiting 4 seconds for /sys/block/sda/dev to show up
Waiting 4 seconds for /sys/block/sda/dev to show up
+ sleep 4
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Waiting 8 seconds for /sys/block/sda/dev to show up
Waiting 8 seconds for /sys/block/sda/dev to show up
+ sleep 8
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Waiting 16 seconds for /sys/block/sda/dev to show up
Waiting 16 seconds for /sys/block/sda/dev to show up
+ sleep 16
+ /bin/cat /sys/block/sda/dev
/bin/cat: /sys/block/sda/dev: No such file or directory
+ devpair=
+ [  =  ]
+ echo Device /sys/block/sda/dev seems to be down.
Device /sys/block/sda/dev seems to be down.
+ [ yes !=  ]
+ echo Debugging opportunity, type ^D to continue.
Debugging opportunity, type ^D to continue.
+ /bin/dash
/bin/dash: can't access tty; job control turned off

At this stage I get a dash prompt at which I manually create /dev/sda, /dev/sda[12] and /dev/sda5 (why does it expect sda5 of all places?)
init fails because /proc/block/sda/dev doesn't exist (there are /dev/block/sda[12]), after I create these devices it still fails on not finding 'xen' volume group
(why does it expect it? It has a raw partition to use):

Device /sys/block/sda/sda5/dev seems to be down.
+ [ yes !=  ]
+ echo Debugging opportunity, type ^D to continue.
Debugging opportunity, type ^D to continue.
+ /bin/dash
/bin/dash: can't access tty; job control turned off
# + maj=
+ min=
+ /bin/mknod /dev/sda5 b
/bin/mknod: missing operand after `b'
Special files require major and minor device numbers.
Try `/bin/mknod --help' for more information.
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/drivers/md/dm- mod.ko
device-mapper: ioctl: 4.7.0-ioctl (2006-06-24) initialised: dm-devel <at> redhat.com
+ [ ! -c /dev/mapper/control ]
+ /bin/mkdir /dev/mapper
/bin/mkdir: cannot create directory `/dev/mapper': File exists
+ mkcdev /dev/mapper/control misc/device-mapper
+ mksymdev /dev/mapper/control /sys/class/misc/device-mapper/dev c
+ devfile=/dev/mapper/control
+ sysfile=/sys/class/misc/device-mapper/dev
+ cb=c
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ /bin/cat /sys/class/misc/device-mapper/dev
+ devpair=10:62
+ [ 10:62 =  ]
+ maj=10
+ min=62
+ /bin/mknod /dev/mapper/control c 10 62
+ /sbin/vgchange -a y xen
  Volume group "xen" not found
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/fs/jbd/jbd.ko
+ /sbin/insmod /lib/modules/2.6.18-xen/kernel/fs/ext3/ext3.ko
+ /bin/mount -n -r -t ext3 -o errors=remount-ro,noatime /dev/mapper/xen-root /mnt
mount: special device /dev/mapper/xen-root does not exist
+ switchroot ydebug text
+ [ yes !=  ]
+ echo Debugging opportunity, ^D to continue.
Debugging opportunity, ^D to continue.
+ /bin/dash
/bin/dash: can't access tty; job control turned off
# + echo Switching root ...
Switching root ...
+ /bin/umount -n /sys
+ /bin/umount -n /proc
+ exec /usr/lib/yaird/exec/run_init /mnt /sbin/init ydebug text
/usr/lib/yaird/exec/run_init: current directory on the same filesystem as the root: Success
Kernel panic - not syncing: Attempted to kill init!


Can anyone point me at what am I doing wrong?

Thanks,

--Amos

_______________________________________________
Xen-users mailing list
Xen-users <at> lists.xensource.com
http://lists.xensource.com/xen-users
Permalink | Reply | 0 comments |
headers
Stefan de Konink | 1 Nov 01:54
Picon
Picon
Favicon

Re: HVM PV drivers lincense with out xen-center


Yung Giang schreef:
>          I was wonder if anyone knows how to input the PV drivers
> license into dom0 with out using xen-center/xen api and with
> xend/xenstore?
> 
> Thank you for your time and help,

Now there is something you should know before trying this out. If you
get it running (aka download the XenServer Express) copy the ISO with PV
drivers and put the iso link in your config file. (This all answers your
question.)

There is a license agreement that you *should not use* this with any
other product. I guess this includes the Open Source version. Since I
thought I could ignore this license agreement, because such things are
not valid in my tiny country, I noticed that the next boot of Windows
included a message that a "service could not be started", and that I had
some PCI Xen device. A SCSI device driver with an exclamation mark and
it all didn't feel better than before.

So... register yourself as Xen Express user you will find the iso no
doubt and see for yourself. You probably want to run Xen Express to, to
make a real judgment. I sincerely hope the KVM guys will make some Open
Source PV drivers that can work with Xen too and otherwise we should
wait for 'Windows Virtualisation' which does everything natively ;)

Stefan
Permalink | Reply | 1 comment |
headers
James Harper | 1 Nov 02:59
Picon

RE: HVM PV drivers lincense with out xen-center

> So... register yourself as Xen Express user you will find the iso no
> doubt and see for yourself. You probably want to run Xen Express to,
to
> make a real judgment. I sincerely hope the KVM guys will make some
Open
> Source PV drivers that can work with Xen too and otherwise we should
> wait for 'Windows Virtualisation' which does everything natively ;)

I've written some PV block device drivers for windows (see recent email
to -dev list). Very very alpha at this stage (eg it barely works), but
it should get more stable pretty rapidy (I think I've found the main
cause of the instability - I'm breaking lots of WinDDK rules about when
I can wait and call certain functions etc)

James
Permalink | Reply | 0 comments |
headers
Tomoki Taniguchi | 1 Nov 09:30
Picon

64bit xen?

we have recently purchased a 2x dual core 64bit Xeon server.
I want to turn it into a xen server.  But i have never played with
64bit version of
xen before.

Is there anything i should be careful with when running
the 64bit version of Xen 3.1?

I have read 3.1 allows me to run 32bit domUs
but is it better to run the domUs as 64bit when ever possible?

TIA,
Tomoki
Permalink | Reply | 2 comments |
headers
Nico Kadel-Garcia | 1 Nov 09:53
Picon

Re: 64bit xen?

Tomoki Taniguchi wrote:
> we have recently purchased a 2x dual core 64bit Xeon server.
> I want to turn it into a xen server.  But i have never played with
> 64bit version of
> xen before.
>
> Is there anything i should be careful with when running
> the 64bit version of Xen 3.1?
>
> I have read 3.1 allows me to run 32bit domUs
> but is it better to run the domUs as 64bit when ever possible?
>
>
> TIA,
> Tomoki
>   
What is your underlying OS? And does your server support full 
virtualization?
Permalink | Reply | 0 comments |
headers
Nico Kadel-Garcia | 1 Nov 10:03
Picon

Re: help for physical partition for booting XenU

techy geek wrote:
> Hi All,
>
> I need help for creating private file system for Guest OS: XenU
>
> I had successfully booted my host OS with Xen0 which is running 
> linux2.6.18 version  and is using the physical device /dev/hda1.
>
> Now my next step is to  boot the guest OS and  for that  I first need 
> to create the private file system for this  guest OS. As you mentioned 
> there could be 3 methods to create a private file system:
> * File Based image
> * LVM Based
> * Physical Partition
>
> Now as per my understanding, File Based Image is easy to create but 
> has some poor I/O performance and from I/O performance perspective, 
> Physical Partition is better. My requirement for installing Xen is  to 
> do some OS profiling on Guest OS, so that way it seems Physical 
> Partition could be right choice for me.
>
> Could you please let me know the step-by-step approach to set up 
> Physical partiton for booting XenU. Also let me know the  changes I 
> need to make for my config file for booting XenU-(Guest OS).
>
> Also, if possible let me know for file based image so that I can have 
> one different domain running on File Based image(File system) and can  
> try profiling on this also.
>
> My available drives are /dex/hda2 /dev/sda1 /dev/sda2 /dev/sdb1 /dev/sdb2
I like using LVM, because I can more gracefully resize or snapshot the 
partitions and work with the snapshot, rather than the original 
partition, for backup purposes. It's performance is similar if not 
identical to that of a physical partition.

Are you comfortable working with fdisk to create partitions? And what 
does your partitioning look like right now? If all your extra disk is 
already in LVM "physical volumes" and "group volumes", it'll be simpler 
to create a new "logical volume" with the command 'lvcreate'.

You've also failed to say what OS your Dom0 is, which affects the tools 
available for building your DomU, and whether you want full 
virtualization or para-virtualization. It makes a difference!
Permalink | Reply | 0 comments |
headers
sadegh | 2 Nov 02:24
Picon

Re: Xen clustering

Tom Brown wrote:
> On Wed, 31 Oct 2007, Nico Kadel-Garcia wrote:
>
>> shacky wrote:
>>>  Hi.
>>>
>>>  I'm writing to the mailing list to know if Xen is what I'm looking 
>>> for.
>>>  I need a failover cluster with two completely and automatically
>>>  replicated and redundanced servers running Windows Server 2003
>>>  Standard.
>>>  Could Xen make this?
>>>  Where I can found some information about making this configuration 
>>> with
>>>  Xen?
>>>
>>>  Thank you very much for your help!
>>>
>>>  Bye.
>>>
>> Not..... exactly. You can keep multiple operating systems on a single 
>> server: you can migrate them from a common network storage to run on 
>> one server or another. But failover where one server's operating 
>> system fails or is disabled, and another takes over the function, is 
>> a much more interesting thing to do. To do complete OS failover, 
>> every physical component must be mirrored and matched and the 
>> connections duplicated
>
> what? Even if "every component must be matched", that would be a GREAT 
> reason to use XEN, as all the components are virtual, so of course 
> they match.
>
> .. And using XEN underneath a proprietory O/S allows you to do things 
> like replicate the filesystem via DRBD to another node... and/or make 
> backups via snapshots, etc...
>
> So, while the question is incomplete, and XEN isn't going to do the 
> job by itself, it is a very usefull tool in getting the job done... as 
> putting the proprietary OS inside a virtual machine gives you a degree 
> of control you might not be able to get on physical hardware. I've got 
> a couple of physical windows servers that I _wish_ were running inside 
> virtual machines.
>
> That said, while I do use DRBD in production, I have not configured 
> heartbeat to configure automated failover of services... nor do I make 
> much use of windows.
>
> -Tom
>
>> with failover redundancy, and that gets fascinating to support. It's 
>> also amazingly expensive, much more than the cost of a small set of 
>> duplicate servers.
>
> Nico, I like some of your answers... and some of them leave my mouth 
> hanging open. You don't leave yourself much wiggle room, it would be 
> really easy for some newbie to think they were "the definitive" answers.
>
>>
>> If you have specific services running on those servers, you might be 
>> able to gracefully implement them in normal master/slave or 
>> master/master failover modes.
>>
>> _______________________________________________
>> Xen-users mailing list
>> Xen-users <at> lists.xensource.com
>> http://lists.xensource.com/xen-users
>>
>>
>
> ----------------------------------------------------------------------
> tbrown <at> BareMetal.com   | Always bear in mind that your own resolution to
> http://BareMetal.com/  | success is more important than any other one
> web hosting since '95  | thing. - Abraham Lincoln
>
> _______________________________________________
> Xen-users mailing list
> Xen-users <at> lists.xensource.com
> http://lists.xensource.com/xen-users
>
>
>
Hi All,
    well , if  we  say  FailOver mechanism is stop one and restart 
another node it is true but if FailOver definition is  Migration from 
one node to another.
What is your idea about it?
increasing redundant hardwares is it?
and if we have proactive fault detection mechanism we can do it.
some related works exist.
My thesis is about this subject "Improving Survivability of  HA 
Clusters" (that Mission-Critical apps. runs on...)
if any body help me to Hacking Xen, heartbeat and implementing some 
other tools and compare results we can  publishing some Papers about 
this and have friendly work group!
I am ready to contribute and very appreciate.
Best Regards
Sadegh
Permalink | Reply | 2 comments |
headers
Arun Sharma | 1 Nov 15:03
Picon

XEN dom0 - mount: could not find filestystem '/dev/root'.

Hi

Error i am getting :

mount: could not find filestystem '/dev/root'.
Setting up other filesystems
Setting up new root fs
setuproot: moving /dev failed: No such file or directory
no fstab.sys mounting internal defaults
setuproot: error mounting /proc: No such file or directory
setuproot: error mounting /sys: No such file or directory
Switching to new root and running init
unmounting old /dev/
unmounting old /proc
unmounting old /sys
switchroot: mount failed: No such file or directory
Kernel panic - not syncing: Attempted to kill init


I have done a source build for Xen 3.1 and created initrd image after a successfull build for dom0.
But it is giving kernel panic when it buits dom0

I have created initrd image using below commands (MIND IT : I HAVE IDE HARD DISK) and tried with each one.. but no help.

mkinitrd -v -f --with=aacraid --with=sd_mod --with=scsi_mod initrd-2.6.18-xen.img 2.6.18-xen
mkinitrd -v -f --force-ide-probe --fstab=/etc/fstab initrd-2.6.18-xen.img 2.6.18-xen
mkinitrd -v -f --omit-scsi-modules initrd-2.6.18-xen.img 2.6.18-xen
mkinitrd -v -f --with=xennet --preload=xenblk initrd-2.6.18-xen.img 2.6.18-xen
 mkinitrd -v -f --force-ide-probe --fstab=/etc/fstab initrd-2.6.18-xen.img 2.6.18-xen

Something similar to these.
http://lists.xensource.com/archives/html/xen-users/2007-04/msg00182.html
--


Regards
Arun Sharma
0-934-829-3810 

_______________________________________________
Xen-users mailing list
Xen-users <at> lists.xensource.com
http://lists.xensource.com/xen-users
Permalink | Reply | 1 comment |
headers
Imre Oolberg | 1 Nov 15:17
Picon

dom0-cpus parameter not taking effect on quad-core intel under debian etch

Hallo!

I have used Xen 3.0.3 packaged with Debian Etch on 64bit platform and
with AMD dual-core Opteron processors successfully for some time,
overcoming the known problem of starting domUs locking down the computer
by using in xend-config.sxp a line

(dom0-cpus 1)

To see if this line takes effect i look after starting Xen under
/proc/cpuinfo and see there only one processor.

Doing the very same configuration with the same software on 2xquad-core
intel processor computer (on HP ProLiant BL 460c G1) i get 5 processors

bash# cat /proc/cpuinfo | grep processor
processor       : 0
processor       : 4
processor       : 5
processor       : 6
processor       : 7

cpuid says that processor is

Extended brand string: "Intel(R) Xeon(R) CPU           X5355  @ 2.66GHz"

If someone could give an explaination i'd be thankful! Its true, 3.0.3
xen is quite old but i care about package management and until now cant
complain about it satifying my needs.

Best regards,

Imre
Permalink | Reply | 0 comments |
headers
Nico Kadel-Garcia | 1 Nov 16:30
Picon

Re: Xen clustering

sadegh wrote:
>
> Hi All,
>    well , if  we  say  FailOver mechanism is stop one and restart 
> another node it is true but if FailOver definition is  Migration from 
> one node to another.
> What is your idea about it?
Slow down a moment. Failover means different things to different people. 
A full so-called "High Availability" solution, where each individual 
component can fail but the rest automatically switch the resources 
they're using, is useful sometimes but seriously expensive to implement, 
and the switchover mechanisms themselves create their own uncertainties.

> increasing redundant hardwares is it?
It's part of an integrated solution. The redundant hardware has to be 
cross-wired in usable failover setups for individual setups. Xen can be 
a very useful component for this, because the potentially more reliable 
centralized storage system can be made extremely robust and the servers 
swap domains as necessary to other working hardware.
> and if we have proactive fault detection mechanism we can do it.
Yes, but there are limits to this. Janitors blow fuses: cage monkeys 
accidentally crimp fiber, or disconnect idle connections. And hard 
drives fail without any warning whatsoever, even a few at a time. 
There's a white paper from Google on this I highly recommend, where 
roughly 30% of the drives fail without any SMART detection at all.

> some related works exist.
> My thesis is about this subject "Improving Survivability of  HA 
> Clusters" (that Mission-Critical apps. runs on...)
> if any body help me to Hacking Xen, heartbeat and implementing some 
> other tools and compare results we can  publishing some Papers about 
> this and have friendly work group!
> I am ready to contribute and very appreciate.
> Best Regards
> Sadegh
Ohhhh. Cool. It's not clear how much you need Xen for this, as much as 
integration of Xen with the available approaches.
Permalink | Reply | 1 comment |

Gmane